• John
  • Felde
  • University of Maryland
  • USA

Latest Posts

  • James
  • Doherty
  • Open University
  • United Kingdom

Latest Posts

  • Andrea
  • Signori
  • Nikhef
  • Netherlands

Latest Posts

  • CERN
  • Geneva
  • Switzerland

Latest Posts

  • Aidan
  • Randle-Conde
  • Université Libre de Bruxelles
  • Belgium

Latest Posts

  • TRIUMF
  • Vancouver, BC
  • Canada

Latest Posts

  • Laura
  • Gladstone
  • MIT
  • USA

Latest Posts

  • Steven
  • Goldfarb
  • University of Michigan

Latest Posts

  • Fermilab
  • Batavia, IL
  • USA

Latest Posts

  • Seth
  • Zenz
  • Imperial College London
  • UK

Latest Posts

  • Nhan
  • Tran
  • Fermilab
  • USA

Latest Posts

  • Alex
  • Millar
  • University of Melbourne
  • Australia

Latest Posts

  • Ken
  • Bloom
  • USLHC
  • USA

Latest Posts

Byron Jennings | TRIUMF | Canada

View Blog | Read Bio

Good Management is Science

Management done properly satisfies Sir Karl Popper’s (1902 – 1994) demarcation criteria for science, i.e. using models that make falsifiable or at least testable predictions. That was brought home to me by a book[1] by Douglas Hubbard on risk management where he advocated observationally constrained (falsifiable or testable) models for risk analysis evaluated through Monte Carlo calculations. Hmm, observationally constrained models and Monte Carlo calculations, sounds like a recipe for science.

Let us take a step back. The essence of science is modeling how the universe works and checking the assumptions of the model and its predictions against observations. The predictions must be testable. According to Hubbard, the essence of risk management is modeling processes and checking the assumptions of the model and its predictions against observations. The predictions must be testable. What we are seeing here is a common paradigm for knowledge in which modeling and testing against observation play a key role.

The knowledge paradigm is the same in project management. A project plan, with its resource loaded schedules and other paraphernalia, is a model for how the project is expected to proceed. To monitor a project you check the plan (model) against actuals (a fancy euphemism for observations, where observations may or may not correspond to reality). Again, it reduces back to observationally constrained models and testable predictions.

The foundations of science and good management practices are tied even closer together. Consider the PDCA cycle for process management that is present, either implicitly or explicitly, in essentially all the ISO standards related to management. It was originated by Walter Shewhart (1891 – 1967), an American physicist, engineer and statistician, and popularized by Edwards Deming (1900 – 1993), an American engineer, statistician, professor, author, lecturer and management consultant. Engineers are into everything. The actual idea of the cycle is based on the ideas of Francis Bacon (1561 – 1629) but could equally well be based on the work of Roger Bacon[2] (1214 – 1294). Hence, it should probably be called the Double Bacon Cycle (no, that sounds too much like a breakfast food).

But what is this cycle? For science, it is: plan an experiment to test a model, do the experiment, check the model results against theCapture observed results, and act to change the model in response to the new information from the check stage or devise more precise tests if the predictions and observations agree. For process management replace experiment with production process. As a result, you have a model for how the production process should work and doing the process allows you to test the model. The check stage is where you see if the process performed as expected and the act stage allows you to improve the process if the model and actuals do not agree. The key point is the check step. It is necessary if you are to improve the process; otherwise you do not know what is going wrong or, indeed, even if something is going wrong. It is only possible if the plan makes predictions that are falsifiable or at least testable. Popper would be pleased.

There is another interesting aspect of the ISO 9001 standard. It is based on the idea of processes. A process is defined as an activity that converts inputs into outputs. Well, that sound rather vague, but the vagueness is an asset, kind of like degrees of freedom in an effective field theory. Define them as you like but if you choose them incorrectly you will be sorry. The real advantage of effective field theory and the flexible definition of process is that you can study a system at any scale you like. In effective field theory, you study processes that operate at the scale of the atom, the scale of the nucleus or the scale of the nucleon and tie them together with a few parameters. Similarly with processes, you can study the whole organization as a process or drill down and look at sub process at any scale you like, for CERN or TRIUMF that would be down to the last magnet. It would not be useful to go further and study accelerator operations at the nucleon scale. At a given scale different processes are tied together by their inputs and outputs and these are also used to tie process at different scales.

As a theoretical physicist who has gone over to the dark side and into administration, I find it amusing to see the techniques and approaches from science being borrowed for use in administration, even Monte Carlo calculations. The use of similar techniques in science and administration goes back to the same underlying idea: all true knowledge is obtained through observation and its use to build better testable models, whether in science or other walks of life.

[1] The Failure of Risk Management: Why It’s Broken and How to Fix It by Douglas W. Hubbard (Apr 27, 2009)

[2] Roger Bacon described a repeating cycle of observation, hypothesis, and experimentation.

Share

Tags: , ,

  • Andre Mirabelli

    Byron,
    As a fellow theoretical physicist who has also shifted focus, I am happy to see a discussion of how modern confirmation theory has expanded its purview. But I did want to indicate an additional step in the discovery and evaluation process you describe. While Popper did not always make it clear, after one checks a model’s predictions against observations one still does not know if one should accept that model until one sees how well it has checked out in comparison to other competing seriously proposed models. That a model tests well is not convincing of its appropriateness if another model of the same material tests much better yet. It is this comparison of models (formally through ‘likelihood ratio tests’) that tells us what model to rely upon now and suggests what changes in past models to look into next.

  • My discussion of Popper here is very sketchy. As you note, he made the point that what you can do is say which of two models is better. The point of the article is that, as with science,good management also requires checking predictions against observations (actuals)and reacting to any differences. It is probably true, that in this case as well, all you can say is which of two approaches is better.